Single omics data analysis
19 février 2021
1 Loading
latmx2.mset <- phenomis::reading(ProMetIS::post_processed_dir.c(),
report.c = "none")
latmx2.mset <- latmx2.mset[, ProMetIS::sets.vc()]2 Splitting LAT and MX2
Discarding features with either:
NAs > 20%
variance < 1e-5
proteomics: imputation > 20% in both conditions
gene_mset.ls <- lapply(ProMetIS::genes.vc(),
function(gene.c) {
message(gene.c)
ProMetIS::subsetting(latmx2.mset,
genes.vc = c("WT", gene.c))
})## LAT## Discarded 33 features: Haematology...Complete.blood.count..CBC.on.the.Advia.120..BASO...., Haematology...Complete.blood.count..CBC.on.the.Advia.120..EOSINO...., Haematology...Complete.blood.count..CBC.on.the.Advia.120..LUC...., Haematology...Complete.blood.count..CBC.on.the.Advia.120..LYMPHO...., Haematology...Complete.blood.count..CBC.on.the.Advia.120..MONO...., Haematology...Complete.blood.count..CBC.on.the.Advia.120..NEUTRO...., Haematology...Complete.blood.count..CBC.on.the.Advia.120.CHCM..g.dL., Haematology...Complete.blood.count..CBC.on.the.Advia.120.CHDW..g.dL., Haematology...Complete.blood.count..CBC.on.the.Advia.120.EOSINO..x10E03.cells.µL., Haematology...Complete.blood.count..CBC.on.the.Advia.120.EOSINO.BRUT..x10E03.cells.µL., Haematology...Complete.blood.count..CBC.on.the.Advia.120.HCT...., Haematology...Complete.blood.count..CBC.on.the.Advia.120.HGB..g.dL., Haematology...Complete.blood.count..CBC.on.the.Advia.120.HGB.brut..g.dL., Haematology...Complete.blood.count..CBC.on.the.Advia.120.Hte.brut...., Haematology...Complete.blood.count..CBC.on.the.Advia.120.LUC..x10E03.cells.µL., Haematology...Complete.blood.count..CBC.on.the.Advia.120.LUC.BRUT..x10E03.cells.µL., Haematology...Complete.blood.count..CBC.on.the.Advia.120.LYMPHO..x10E03.cells.µL., Haematology...Complete.blood.count..CBC.on.the.Advia.120.LYMPHO.BRUT..x10E03.cells.µL., Haematology...Complete.blood.count..CBC.on.the.Advia.120.MCH..pg., Haematology...Complete.blood.count..CBC.on.the.Advia.120.MCHC..g.dL., Haematology...Complete.blood.count..CBC.on.the.Advia.120.MCV..fL., Haematology...Complete.blood.count..CBC.on.the.Advia.120.MONO..x10E03.cells.µL., Haematology...Complete.blood.count..CBC.on.the.Advia.120.MONO.BRUT..x10E03.cells.µL., Haematology...Complete.blood.count..CBC.on.the.Advia.120.MPV..fL., Haematology...Complete.blood.count..CBC.on.the.Advia.120.NEUTRO..x10E03.cells.µL., Haematology...Complete.blood.count..CBC.on.the.Advia.120.NEUTRO.BRUT..x10E03.cells.µL., Haematology...Complete.blood.count..CBC.on.the.Advia.120.PLT..x10E03.cells.µL., Haematology...Complete.blood.count..CBC.on.the.Advia.120.PLT.brut..x10E03.cells.µL., Haematology...Complete.blood.count..CBC.on.the.Advia.120.RBC..x10E06.cells.µL., Haematology...Complete.blood.count..CBC.on.the.Advia.120.RBC.brut..x10E06.cells.µL., Haematology...Complete.blood.count..CBC.on.the.Advia.120.RDW...., Haematology...Complete.blood.count..CBC.on.the.Advia.120.WBC..x10E03.cells.µL., Haematology...Complete.blood.count..CBC.on.the.Advia.120.WBC.brut..x10E03.cells.µL.## Nb of discard. feat. in 'preclinical': nas_zerovar: 33, overimputed: 0## Nb of discard. feat. in 'proteomics_liver': nas_zerovar: 0, overimputed: 89## Nb of discard. feat. in 'proteomics_plasma': nas_zerovar: 0, overimputed: 27## Nb of discard. feat. in 'metabolomics_liver_c18hypersil_pos': nas_zerovar: 0, overimputed: 0## Nb of discard. feat. in 'metabolomics_liver_hilic_neg': nas_zerovar: 0, overimputed: 0## Nb of discard. feat. in 'metabolomics_plasma_c18hypersil_pos': nas_zerovar: 0, overimputed: 0## Nb of discard. feat. in 'metabolomics_plasma_hilic_neg': nas_zerovar: 0, overimputed: 0## Nb of discard. feat. in 'metabolomics_plasma_c18acquity_pos': nas_zerovar: 0, overimputed: 0## Nb of discard. feat. in 'metabolomics_plasma_c18acquity_neg': nas_zerovar: 0, overimputed: 0## MX2## Discarded 2 features: Eye.Morphology.OCT.right.anterior.chamber.depth, Eye.Morphology.OCT.right.corneal.thickness## Discarded 3 samples: W621f, W623f, W633f## Nb of discard. feat. in 'preclinical': nas_zerovar: 2, overimputed: 0## Nb of discard. feat. in 'proteomics_liver': nas_zerovar: 0, overimputed: 97## Nb of discard. feat. in 'proteomics_plasma': nas_zerovar: 0, overimputed: 24## Nb of discard. feat. in 'metabolomics_liver_c18hypersil_pos': nas_zerovar: 0, overimputed: 0## Nb of discard. feat. in 'metabolomics_liver_hilic_neg': nas_zerovar: 0, overimputed: 0## Nb of discard. feat. in 'metabolomics_plasma_c18hypersil_pos': nas_zerovar: 0, overimputed: 0## Nb of discard. feat. in 'metabolomics_plasma_hilic_neg': nas_zerovar: 0, overimputed: 0## Nb of discard. feat. in 'metabolomics_plasma_c18acquity_pos': nas_zerovar: 0, overimputed: 0## Nb of discard. feat. in 'metabolomics_plasma_c18acquity_neg': nas_zerovar: 0, overimputed: 0names(gene_mset.ls) <- ProMetIS::genes.vc()3 Univariate hypothesis testing
for (gene.c in ProMetIS::genes.vc()) {
message(gene.c)
gene.mset <- gene_mset.ls[[gene.c]]
if (gene.c == "MX2") {
# all sets: performing 'limma' 2 ways testing for gene and sex
gene.mset <- phenomis::hypotesting(gene.mset,
test.c = "limma2ways",
factor_names.vc = c("gene", "sex"),
factor_levels.ls = list(factor1.vc = c("WT", gene.c),
factor2.vc = ProMetIS::sex.vc()),
signif_maxprint.i = 10,
title.c = gene.c,
report.c = "none")
} else if (gene.c == "LAT") {
# 'proteomics_liver' set: performing the 'limma' testing for gene in the male and female subsets, and 'limma' testing for sex in the LAT and WT subsets
# all other sets: performing 'limma' 2 ways testing for gene and sex
protliv.eset <- gene.mset[["proteomics_liver"]]
protliv_fda.df <- Biobase::fData(protliv.eset)
## 'proteomics_liver': 'limma' testing for gene in the male and female subsets
for (sex.c in ProMetIS::sex.vc()) {
protlivsex.eset <- ProMetIS::subsetting(protliv.eset,
set.c = "proteomics_liver",
genes.vc = c("WT", "LAT"),
sex.vc = sex.c)
protlivsex.eset <- phenomis::hypotesting(protlivsex.eset,
test.c = "limma",
factor_names.vc = "gene",
factor_levels.ls = list(factor1.vc = c("WT", gene.c)),
signif_maxprint.i = 10,
title.c = paste0("proteomics_liver, ", sex.c),
report.c = "none")
protlivsex.df <- Biobase::fData(protlivsex.eset)
limmasex.df <- protlivsex.df[, grep("limma", colnames(protlivsex.df))]
colnames(limmasex.df) <- gsub("limma_gene_",
paste0("limma", sex.c, "_"),
colnames(limmasex.df))
protliv_fda.df <- merge(protliv_fda.df,
limmasex.df,
by = 0, all = TRUE, sort = FALSE)
rownames(protliv_fda.df) <- protliv_fda.df[, "Row.names"]
protliv_fda.df[, "Row.names"] <- NULL
}
stopifnot(identical(sort(rownames(protliv_fda.df)),
sort(Biobase::featureNames(protliv.eset))))
Biobase::fData(protliv.eset) <- protliv_fda.df[Biobase::featureNames(protliv.eset), ]
## 'proteomics_liver': 'limma' testing for sex in the LAT and WT subsets
for (gene.c in c("WT", "LAT")) {
protlivgene.eset <- protliv.eset[, Biobase::pData(protliv.eset)[, "gene"] == gene.c]
protlivgene.eset <- ProMetIS::subsetting(protlivgene.eset,
set.c = "proteomics_liver",
genes.vc = gene.c,
sex.vc = ProMetIS::sex.vc())
protlivgene.eset <- phenomis::hypotesting(protlivgene.eset,
test.c = "limma",
factor_names.vc = "sex",
factor_levels.ls = list(factor1.vc = ProMetIS::sex.vc()),
signif_maxprint.i = 10,
title.c = paste0("proteomics_liver, ", gene.c),
report.c = "none")
protlivgene.df <- Biobase::fData(protlivgene.eset)
limmagene.df <- protlivgene.df[, grep("limma_sex_",
colnames(protlivgene.df))]
colnames(limmagene.df) <- gsub("limma_sex_",
paste0("limma", gene.c, "_"),
colnames(limmagene.df))
protliv_fda.df <- merge(protliv_fda.df,
limmagene.df,
by = 0, all = TRUE, sort = FALSE)
rownames(protliv_fda.df) <- protliv_fda.df[, "Row.names"]
protliv_fda.df[, "Row.names"] <- NULL
}
stopifnot(identical(sort(rownames(protliv_fda.df)),
sort(Biobase::featureNames(protliv.eset))))
Biobase::fData(protliv.eset) <- protliv_fda.df[Biobase::featureNames(protliv.eset), ]
## all other sets: 'limma2ways' testing for gene and sex
gene.mset <- gene.mset[, setdiff(names(gene.mset), "proteomics_liver")]
gene.mset <- phenomis::hypotesting(gene.mset,
test.c = "limma2ways",
factor_names.vc = c("gene", "sex"),
factor_levels.ls = list(factor1.vc = c("WT", gene.c),
factor2.vc = ProMetIS::sex.vc()),
signif_maxprint.i = 10,
title.c = gene.c,
report.c = "none")
# including the 'proteomics_liver' dataset back
gene.mset <- MultiDataSet::add_eset(gene.mset,
protliv.eset,
dataset.type = "proteomics_liver",
GRanges = NA,
overwrite = TRUE,
warnings = FALSE)
# re-ordering
gene.mset <- gene.mset[,
ProMetIS::sets.vc()[ProMetIS::sets.vc() %in% names(gene.mset)]]
}
gene_mset.ls[[gene.c]] <- gene.mset
}## LAT## Nb of discard. feat. in 'proteomics_liver': nas_zerovar: 0, overimputed: 14## Nb of discard. feat. in 'proteomics_liver': nas_zerovar: 0, overimputed: 4
## Nb of discard. feat. in 'proteomics_liver': nas_zerovar: 0, overimputed: 20
## Nb of discard. feat. in 'proteomics_liver': nas_zerovar: 0, overimputed: 5
## MX2
4 Principal Component Analysis
Score plot colored according to genotype or sex.
for (gene.c in ProMetIS::genes.vc()) {
message(gene.c)
gene.mset <- gene_mset.ls[[gene.c]]
gene_mset.pca <- ropls::opls(gene.mset, fig.pdfC = "none")
ropls::plot(gene_mset.pca, plotPhenoDataC = "gene", typeVc = "x-score",
parPaletteVc = ProMetIS::palette.vc()[rev(c("WT", gene.c))])
ropls::plot(gene_mset.pca, plotPhenoDataC = "sex", typeVc = "x-score",
parPaletteVc = ProMetIS::palette.vc()[rev(ProMetIS::sex.vc())])
gene.mset <- ropls::getMset(gene_mset.pca)
gene_mset.ls[[gene.c]] <- gene.mset
}## LAT##
##
## Building the model for the 'preclinical' dataset:
## PCA
## 28 samples x 203 variables
## standard scaling of predictors
## 245 (4%) NAs
## R2X(cum) pre ort
## Total 0.504 4 0
##
##
## Building the model for the 'proteomics_liver' dataset:
## PCA
## 28 samples x 2098 variables
## standard scaling of predictors
## R2X(cum) pre ort
## Total 0.51 5 0
##
##
## Building the model for the 'proteomics_plasma' dataset:
## PCA
## 24 samples x 419 variables
## standard scaling of predictors
## R2X(cum) pre ort
## Total 0.535 4 0
##
##
## Building the model for the 'metabolomics_liver_c18hypersil_pos' dataset:
## PCA
## 28 samples x 5665 variables
## standard scaling of predictors
## R2X(cum) pre ort
## Total 0.548 4 0
##
##
## Building the model for the 'metabolomics_liver_hilic_neg' dataset:
## PCA
## 28 samples x 2866 variables
## standard scaling of predictors
## R2X(cum) pre ort
## Total 0.545 4 0
##
##
## Building the model for the 'metabolomics_plasma_c18hypersil_pos' dataset:
## PCA
## 28 samples x 4788 variables
## standard scaling of predictors
## 1 (0%) NAs
## R2X(cum) pre ort
## Total 0.525 5 0
##
##
## Building the model for the 'metabolomics_plasma_hilic_neg' dataset:
## PCA
## 28 samples x 3131 variables
## standard scaling of predictors
## R2X(cum) pre ort
## Total 0.529 5 0
##
##
## Building the model for the 'metabolomics_plasma_c18acquity_pos' dataset:
## PCA
## 28 samples x 6104 variables
## standard scaling of predictors
## 76 (0%) NAs
## R2X(cum) pre ort
## Total 0.525 6 0
##
##
## Building the model for the 'metabolomics_plasma_c18acquity_neg' dataset:
## PCA
## 28 samples x 1584 variables
## standard scaling of predictors
## R2X(cum) pre ort
## Total 0.544 6 0
## MX2
##
##
## Building the model for the 'preclinical' dataset:
## PCA
## 29 samples x 234 variables
## standard scaling of predictors
## 407 (6%) NAs
## R2X(cum) pre ort
## Total 0.538 5 0
##
##
## Building the model for the 'proteomics_liver' dataset:
## PCA
## 29 samples x 2090 variables
## standard scaling of predictors
## R2X(cum) pre ort
## Total 0.514 5 0
##
##
## Building the model for the 'proteomics_plasma' dataset:
## PCA
## 25 samples x 422 variables
## standard scaling of predictors
## R2X(cum) pre ort
## Total 0.512 4 0
##
##
## Building the model for the 'metabolomics_liver_c18hypersil_pos' dataset:
## PCA
## 29 samples x 5665 variables
## standard scaling of predictors
## R2X(cum) pre ort
## Total 0.559 5 0
##
##
## Building the model for the 'metabolomics_liver_hilic_neg' dataset:
## PCA
## 29 samples x 2866 variables
## standard scaling of predictors
## R2X(cum) pre ort
## Total 0.55 5 0
##
##
## Building the model for the 'metabolomics_plasma_c18hypersil_pos' dataset:
## PCA
## 29 samples x 4788 variables
## standard scaling of predictors
## 1 (0%) NAs
## R2X(cum) pre ort
## Total 0.508 5 0
##
##
## Building the model for the 'metabolomics_plasma_hilic_neg' dataset:
## PCA
## 29 samples x 3131 variables
## standard scaling of predictors
## R2X(cum) pre ort
## Total 0.546 5 0
##
##
## Building the model for the 'metabolomics_plasma_c18acquity_pos' dataset:
## PCA
## 29 samples x 6104 variables
## standard scaling of predictors
## 77 (0%) NAs
## R2X(cum) pre ort
## Total 0.535 6 0
##
##
## Building the model for the 'metabolomics_plasma_c18acquity_neg' dataset:
## PCA
## 29 samples x 1584 variables
## standard scaling of predictors
## R2X(cum) pre ort
## Total 0.544 6 0
5 OPLS-DA
for (gene.c in ProMetIS::genes.vc()) {
message(gene.c)
gene.mset <- gene_mset.ls[[gene.c]]
gene_mset.oplsda <- ropls::opls(gene.mset, "gene", predI = 1, orthoI = 1,
fig.pdfC = "none")
ropls::plot(gene_mset.oplsda, typeVc = "permutation")
ropls::plot(gene_mset.oplsda, typeVc = "x-score",
parPaletteVc = ProMetIS::palette.vc()[rev(c("WT", gene.c))])
gene.mset <- ropls::getMset(gene_mset.oplsda)
gene_mset.ls[[gene.c]] <- gene.mset
}## LAT##
##
## Building the model for the 'preclinical' dataset:
## OPLS-DA
## 28 samples x 203 variables and 1 response
## standard scaling of predictors and response(s)
## 245 (4%) NAs
## R2X(cum) R2Y(cum) Q2(cum) RMSEE pre ort pR2Y pQ2
## Total 0.161 0.834 0.268 0.215 1 1 0.1 0.1
## No model was included for the 'preclinical' dataset because pQ2 was above 5%.
##
##
## Building the model for the 'proteomics_liver' dataset:
## OPLS-DA
## 28 samples x 2098 variables and 1 response
## standard scaling of predictors and response(s)
## R2X(cum) R2Y(cum) Q2(cum) RMSEE pre ort pR2Y pQ2
## Total 0.273 0.876 0.603 0.186 1 1 0.25 0.05
##
##
## Building the model for the 'proteomics_plasma' dataset:
## OPLS-DA
## 24 samples x 419 variables and 1 response
## standard scaling of predictors and response(s)
## R2X(cum) R2Y(cum) Q2(cum) RMSEE pre ort pR2Y pQ2
## Total 0.312 0.915 0.527 0.155 1 1 0.05 0.05
##
##
## Building the model for the 'metabolomics_liver_c18hypersil_pos' dataset:
## OPLS-DA
## 28 samples x 5665 variables and 1 response
## standard scaling of predictors and response(s)
## R2X(cum) R2Y(cum) Q2(cum) RMSEE pre ort pR2Y pQ2
## Total 0.279 0.937 0.843 0.132 1 1 0.05 0.05
##
##
## Building the model for the 'metabolomics_liver_hilic_neg' dataset:
## OPLS-DA
## 28 samples x 2866 variables and 1 response
## standard scaling of predictors and response(s)
## R2X(cum) R2Y(cum) Q2(cum) RMSEE pre ort pR2Y pQ2
## Total 0.341 0.921 0.826 0.148 1 1 0.05 0.05
##
##
## Building the model for the 'metabolomics_plasma_c18hypersil_pos' dataset:
## OPLS-DA
## 28 samples x 4788 variables and 1 response
## standard scaling of predictors and response(s)
## 1 (0%) NAs
## R2X(cum) R2Y(cum) Q2(cum) RMSEE pre ort pR2Y pQ2
## Total 0.205 0.909 0.435 0.159 1 1 0.05 0.05
##
##
## Building the model for the 'metabolomics_plasma_hilic_neg' dataset:
## OPLS-DA
## 28 samples x 3131 variables and 1 response
## standard scaling of predictors and response(s)
## R2X(cum) R2Y(cum) Q2(cum) RMSEE pre ort pR2Y pQ2
## Total 0.203 0.948 0.599 0.121 1 1 0.05 0.05
##
##
## Building the model for the 'metabolomics_plasma_c18acquity_pos' dataset:
## OPLS-DA
## 28 samples x 6104 variables and 1 response
## standard scaling of predictors and response(s)
## 76 (0%) NAs
## R2X(cum) R2Y(cum) Q2(cum) RMSEE pre ort pR2Y pQ2
## Total 0.128 0.973 0.51 0.0863 1 1 0.05 0.05
##
##
## Building the model for the 'metabolomics_plasma_c18acquity_neg' dataset:
## OPLS-DA
## 28 samples x 1584 variables and 1 response
## standard scaling of predictors and response(s)
## R2X(cum) R2Y(cum) Q2(cum) RMSEE pre ort pR2Y pQ2
## Total 0.179 0.91 0.384 0.158 1 1 0.3 0.05
## No model has been built for the 'preclinical' dataset and thus no plot can be displayed.
## No model has been built for the 'preclinical' dataset and thus no plot can be displayed.
## MX2
##
##
## Building the model for the 'preclinical' dataset:
## OPLS-DA
## 29 samples x 234 variables and 1 response
## standard scaling of predictors and response(s)
## 407 (6%) NAs
## R2X(cum) R2Y(cum) Q2(cum) RMSEE pre ort pR2Y pQ2
## Total 0.219 0.839 0.264 0.212 1 1 0.1 0.1
## No model was included for the 'preclinical' dataset because pQ2 was above 5%.
##
##
## Building the model for the 'proteomics_liver' dataset:
## OPLS-DA
## 29 samples x 2090 variables and 1 response
## standard scaling of predictors and response(s)
## R2X(cum) R2Y(cum) Q2(cum) RMSEE pre ort pR2Y pQ2
## Total 0.307 0.933 0.746 0.137 1 1 0.1 0.05
##
##
## Building the model for the 'proteomics_plasma' dataset:
## OPLS-DA
## 25 samples x 422 variables and 1 response
## standard scaling of predictors and response(s)
## R2X(cum) R2Y(cum) Q2(cum) RMSEE pre ort pR2Y pQ2
## Total 0.286 0.904 0.577 0.165 1 1 0.05 0.05
##
##
## Building the model for the 'metabolomics_liver_c18hypersil_pos' dataset:
## OPLS-DA
## 29 samples x 5665 variables and 1 response
## standard scaling of predictors and response(s)
## R2X(cum) R2Y(cum) Q2(cum) RMSEE pre ort pR2Y pQ2
## Total 0.264 0.832 0.515 0.216 1 1 0.1 0.05
##
##
## Building the model for the 'metabolomics_liver_hilic_neg' dataset:
## OPLS-DA
## 29 samples x 2866 variables and 1 response
## standard scaling of predictors and response(s)
## R2X(cum) R2Y(cum) Q2(cum) RMSEE pre ort pR2Y pQ2
## Total 0.166 0.918 0.542 0.151 1 1 0.05 0.05
##
##
## Building the model for the 'metabolomics_plasma_c18hypersil_pos' dataset:
## OPLS-DA
## 29 samples x 4788 variables and 1 response
## standard scaling of predictors and response(s)
## 1 (0%) NAs
## R2X(cum) R2Y(cum) Q2(cum) RMSEE pre ort pR2Y pQ2
## Total 0.153 0.925 0.495 0.145 1 1 0.05 0.05
##
##
## Building the model for the 'metabolomics_plasma_hilic_neg' dataset:
## OPLS-DA
## 29 samples x 3131 variables and 1 response
## standard scaling of predictors and response(s)
## R2X(cum) R2Y(cum) Q2(cum) RMSEE pre ort pR2Y pQ2
## Total 0.234 0.919 0.49 0.15 1 1 0.05 0.05
##
##
## Building the model for the 'metabolomics_plasma_c18acquity_pos' dataset:
## OPLS-DA
## 29 samples x 6104 variables and 1 response
## standard scaling of predictors and response(s)
## 77 (0%) NAs
## R2X(cum) R2Y(cum) Q2(cum) RMSEE pre ort pR2Y pQ2
## Total 0.161 0.935 0.44 0.134 1 1 0.05 0.05
##
##
## Building the model for the 'metabolomics_plasma_c18acquity_neg' dataset:
## OPLS-DA
## 29 samples x 1584 variables and 1 response
## standard scaling of predictors and response(s)
## R2X(cum) R2Y(cum) Q2(cum) RMSEE pre ort pR2Y pQ2
## Total 0.144 0.909 0.289 0.16 1 1 0.2 0.05
## No model has been built for the 'preclinical' dataset and thus no plot can be displayed.
## No model has been built for the 'preclinical' dataset and thus no plot can be displayed.
6 Feature selection
for (gene.c in ProMetIS::genes.vc()) {
message(gene.c)
gene.mset <- gene_mset.ls[[gene.c]]
gene_mset.biosign <- biosigner::biosign(gene.mset,
"gene",
seedI = 123,
plotTierMaxC = "A")
gene.mset <- biosigner::getMset(gene_mset.biosign)
gene_mset.ls[[gene.c]] <- gene.mset
}## LAT##
##
## Selecting the features for the 'preclinical' dataset:
##
##
## Selecting the features for the 'proteomics_liver' dataset:
## Significant features from 'S' groups:
## plsda randomforest svm
## P62962_Profilin-1 "B" "S" "E"
## P62918_60S ribosomal protein L8 "B" "S" "E"
## P62830_60S ribosomal protein L23 "B" "S" "E"
## P35979_60S ribosomal protein L12 "S" "E" "E"
## Accuracy:
## plsda randomforest svm
## Full 0.860 0.829 0.851
## AS 0.905 0.961 0.817
## S 0.891 0.961 NA
##
##
## Selecting the features for the 'proteomics_plasma' dataset:
## Significant features from 'S' groups:
## plsda randomforest svm
## Q9QZ39_Alpha-N-acetylgalactosam. "S" "S" "S"
## Accuracy:
## plsda randomforest svm
## Full 0.870 0.872 0.888
## AS 0.965 0.972 0.927
## S 0.965 0.972 0.969
##
##
## Selecting the features for the 'metabolomics_liver_c18hypersil_pos' dataset:
## Significant features from 'S' groups:
## plsda randomforest svm
## M420.8761T4.64 "E" "S" "A"
## M426.7409T4.13 "C" "S" "E"
## M351.6679T1.45 "E" "E" "S"
## M613.0296T5.64 "E" "E" "S"
## M613.3621T5.59 "S" "E" "E"
## Accuracy:
## plsda randomforest svm
## Full 0.988 0.998 0.988
## AS 0.997 1.000 1.000
## S 0.974 0.995 1.000
##
##
## Selecting the features for the 'metabolomics_liver_hilic_neg' dataset:
## Significant features from 'S' groups:
## plsda randomforest svm
## M698.0528T10.35 "S" "S" "E"
## M439.1412T2.29 "E" "E" "S"
## Accuracy:
## plsda randomforest svm
## Full 0.988 0.956 0.996
## AS 0.965 0.997 0.967
## S 0.988 0.993 0.974
##
##
## Selecting the features for the 'metabolomics_plasma_c18hypersil_pos' dataset:
## Significant features from 'S' groups:
## plsda randomforest svm
## M340.1039T5.35 "C" "S" "S"
## M645.3722T16.46 "D" "S" "B"
## M250.0387T1.05 "B" "S" "E"
## Accuracy:
## plsda randomforest svm
## Full NA 0.646 0.764
## AS 0.825 0.963 0.819
## S NA 0.948 0.817
##
##
## Selecting the features for the 'metabolomics_plasma_hilic_neg' dataset:
## Significant features from 'S' groups:
## plsda randomforest svm
## M239.0025T2.44 "A" "E" "S"
## M248.0239T10.65 "B" "S" "E"
## M112.0123T10.19 "S" "E" "E"
## M542.739T17.12 "S" "E" "E"
## M145T3.3_Monomethyl.glutaric.acid. "S" "E" "E"
## Accuracy:
## plsda randomforest svm
## Full 0.930 0.860 0.931
## AS 0.952 0.947 0.848
## S 0.994 0.915 0.920
##
##
## Selecting the features for the 'metabolomics_plasma_c18acquity_pos' dataset:
## Significant features from 'S' groups:
## plsda randomforest svm
## M333.9622T1.15 "S" "S" "E"
## Accuracy:
## plsda randomforest svm
## Full NA 0.642 0.772
## AS 0.932 0.860 0.790
## S 0.894 0.839 NA
##
##
## Selecting the features for the 'metabolomics_plasma_c18acquity_neg' dataset:
## Significant features from 'S' groups:
## plsda randomforest svm
## M130.0661T10.3 "S" "S" "A"
## Accuracy:
## plsda randomforest svm
## Full 0.717 0.572 0.719
## AS 0.895 0.786 0.904
## S 0.816 0.788 NA
## MX2
##
##
## Selecting the features for the 'preclinical' dataset:
##
##
## Selecting the features for the 'proteomics_liver' dataset:
## Significant features from 'S' groups:
## plsda randomforest svm
## P62852_40S ribosomal protein S25 "S" "S" "E"
## Q64462_Cytochrome P450 4B1 "S" "E" "B"
## P12791_Cytochrome P450 2B10 "E" "E" "S"
## O55071_Cytochrome P450 2B19 "E" "E" "S"
## Q8R1S9_Sodium-coupled neutral a. "E" "E" "S"
## Accuracy:
## plsda randomforest svm
## Full 0.957 0.940 0.964
## AS 0.975 0.993 0.891
## S 1.000 0.986 0.925
##
##
## Selecting the features for the 'proteomics_plasma' dataset:
## Significant features from 'S' groups:
## plsda randomforest svm
## Q9QZ39_Alpha-N-acetylgalactosam. "S" "S" "S"
## Accuracy:
## plsda randomforest svm
## Full 0.858 0.863 0.854
## AS 0.963 0.946 0.970
## S 0.963 0.946 0.962
##
##
## Selecting the features for the 'metabolomics_liver_c18hypersil_pos' dataset:
## Significant features from 'S' groups:
## plsda randomforest svm
## M243.089T6.43 "S" "S" "E"
## Accuracy:
## plsda randomforest svm
## Full 0.763 0.728 0.766
## AS 0.929 0.914 0.655
## S 0.903 0.914 NA
##
##
## Selecting the features for the 'metabolomics_liver_hilic_neg' dataset:
## Significant features from 'S' groups:
## plsda randomforest svm
## M242.0134T1.57 "S" "S" "S"
## Accuracy:
## plsda randomforest svm
## Full 0.796 0.769 0.794
## AS 0.927 0.894 0.886
## S 0.911 0.868 0.891
##
##
## Selecting the features for the 'metabolomics_plasma_c18hypersil_pos' dataset:
## Significant features from 'S' groups:
## plsda randomforest svm
## M340.1039T5.35 "S" "E" "S"
## M580.6222T5.04 "B" "S" "E"
## Accuracy:
## plsda randomforest svm
## Full NA 0.825 0.797
## AS 0.942 0.947 0.858
## S 0.844 0.943 0.892
##
##
## Selecting the features for the 'metabolomics_plasma_hilic_neg' dataset:
## Significant features from 'S' groups:
## plsda randomforest svm
## M242.0135T1.57 "S" "B" "S"
## M898.4079T8.77 "A" "S" "E"
## Accuracy:
## plsda randomforest svm
## Full 0.836 0.809 0.843
## AS 0.931 0.939 0.929
## S 0.911 0.851 0.901
##
##
## Selecting the features for the 'metabolomics_plasma_c18acquity_pos' dataset:
## Significant features from 'S' groups:
## plsda randomforest svm
## M340.103T8.44 "S" "S" "A"
## M290.1752T8.52 "A" "E" "S"
## Accuracy:
## plsda randomforest svm
## Full 0.722 0.815 0.723
## AS 0.944 0.957 0.927
## S 0.898 0.831 0.778
##
##
## Selecting the features for the 'metabolomics_plasma_c18acquity_neg' dataset:
## Significant features from 'S' groups:
## plsda randomforest svm
## M143.0462T1 "A" "E" "S"
## M129.0193T0.9 "S" "E" "E"
## M237.186T16.3 "E" "E" "S"
## M826.9307T0.8_1 "E" "E" "S"
## Accuracy:
## plsda randomforest svm
## Full 0.641 0.556 0.644
## AS 0.906 0.664 0.793
## S 0.739 NA 0.899
7 Combining
# Merging LAT and MX2 results
# Common column names which have to be individualized
common_fvar.vc <- c("limma2ways_sex_M.F_",
# "limma2waysInter_sex_M.F_",
# "limma2waysInter_gene:sex_",
# "anova2ways_sex_M.F_",
# "anova2waysInter_sex_M.F_",
# "anova2waysInter_gene:sex_",
"limmaWT_M.F_", # proteomics_liver
# "limma_sex_M.F_diff_WT",
# "limma_sex_M.F_BH_WT",
# "limma_sex_M.F_signif_WT",
# "limma_sex_M.F_",
"PCA_xload-p",
"hclust",
"gene_OPLSDA_",
"gene_biosign_")
for (set.c in names(latmx2.mset)) {
# initial ExpressionSet
eset <- latmx2.mset[[set.c]]
# initial fData
fdata.df <- Biobase::fData(eset)
# initial features
features.vc <- Biobase::featureNames(eset)
for (gene.c in ProMetIS::genes.vc()) {
if (set.c %in% names(gene_mset.ls[[gene.c]])) {
# gene fData
gene_fdata.df <- Biobase::fData(gene_mset.ls[[gene.c]][[set.c]])
# adding a 'LAT' or 'MX2' tag at the end of columns with identical names
# for the two 'gene-specific' analyzes
gene_fvar.vc <- colnames(gene_fdata.df)
for (fvar.c in common_fvar.vc) {
common_fvar.vi <- grep(fvar.c, gene_fvar.vc, fixed = TRUE)
if (length(common_fvar.vi)) {
gene_fvar.vc[common_fvar.vi] <- paste0(gene_fvar.vc[common_fvar.vi],
"_", gene.c)
}
}
colnames(gene_fdata.df) <- gene_fvar.vc
# additional name simplification
colnames(gene_fdata.df) <- gsub("gene_biosign_",
"biosign_",
gsub("gene_OPLSDA_",
"OPLSDA_",
gsub("limma2ways_gene_",
"limma2ways_",
gsub("limma2ways_sex_",
"limma2ways_",
colnames(gene_fdata.df)))))
# merging
fdata.df <- merge(fdata.df,
gene_fdata.df[, setdiff(colnames(gene_fdata.df),
colnames(fdata.df))],
by = 0, all = TRUE, sort = FALSE)
rownames(fdata.df) <- fdata.df[, "Row.names"]
fdata.df[, "Row.names"] <- NULL
fdata.df <- fdata.df[features.vc, ]
}
}
Biobase::fData(eset) <- fdata.df
latmx2.mset <- MultiDataSet::add_eset(latmx2.mset,
eset,
dataset.type = set.c,
GRanges = NA,
overwrite = TRUE,
warnings = FALSE)
}8 Re-ordering metadata
latmx2.mset <- ProMetIS:::metadata_select(latmx2.mset,
step.c = "3_statistics_singleomics")## Supplementary metadata written in:
## ../inst/extdata/3_statistics_singleomics/metadata_supp.rdata9 Saving (not run)
phenomis::writing(latmx2.mset,
gsub(ProMetIS::data_dir.c(),
"../../ProMetIS/inst/extdata",
ProMetIS::statistics_singleomics_dir.c()),
overwrite.l = TRUE)## Writing the 'preclinical' dataset...## The following file(s) have been written:
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/preclinical/dataMatrix.tsv
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/preclinical/sampleMetadata.tsv
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/preclinical/variableMetadata.tsv## Writing the 'proteomics_liver' dataset...## The following file(s) have been written:
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/proteomics_liver/dataMatrix.tsv
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/proteomics_liver/sampleMetadata.tsv
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/proteomics_liver/variableMetadata.tsv## Writing the 'proteomics_plasma' dataset...## The following file(s) have been written:
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/proteomics_plasma/dataMatrix.tsv
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/proteomics_plasma/sampleMetadata.tsv
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/proteomics_plasma/variableMetadata.tsv## Writing the 'metabolomics_liver_c18hypersil_pos' dataset...## The following file(s) have been written:
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/metabolomics_liver_c18hypersil_pos/dataMatrix.tsv
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/metabolomics_liver_c18hypersil_pos/sampleMetadata.tsv
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/metabolomics_liver_c18hypersil_pos/variableMetadata.tsv## Writing the 'metabolomics_liver_hilic_neg' dataset...## The following file(s) have been written:
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/metabolomics_liver_hilic_neg/dataMatrix.tsv
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/metabolomics_liver_hilic_neg/sampleMetadata.tsv
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/metabolomics_liver_hilic_neg/variableMetadata.tsv## Writing the 'metabolomics_plasma_c18hypersil_pos' dataset...## The following file(s) have been written:
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/metabolomics_plasma_c18hypersil_pos/dataMatrix.tsv
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/metabolomics_plasma_c18hypersil_pos/sampleMetadata.tsv
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/metabolomics_plasma_c18hypersil_pos/variableMetadata.tsv## Writing the 'metabolomics_plasma_hilic_neg' dataset...## The following file(s) have been written:
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/metabolomics_plasma_hilic_neg/dataMatrix.tsv
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/metabolomics_plasma_hilic_neg/sampleMetadata.tsv
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/metabolomics_plasma_hilic_neg/variableMetadata.tsv## Writing the 'metabolomics_plasma_c18acquity_pos' dataset...## The following file(s) have been written:
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/metabolomics_plasma_c18acquity_pos/dataMatrix.tsv
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/metabolomics_plasma_c18acquity_pos/sampleMetadata.tsv
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/metabolomics_plasma_c18acquity_pos/variableMetadata.tsv## Writing the 'metabolomics_plasma_c18acquity_neg' dataset...## The following file(s) have been written:
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/metabolomics_plasma_c18acquity_neg/dataMatrix.tsv
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/metabolomics_plasma_c18acquity_neg/sampleMetadata.tsv
## ../../ProMetIS/inst/extdata/3_statistics_singleomics/metabolomics_plasma_c18acquity_neg/variableMetadata.tsv## The subfolders have been written in the directory:
## ../../ProMetIS/inst/extdata/3_statistics_singleomics